Chirality recognition of amino acid by combining machine learning method and sliding window technique

Chiral molecule recognition is crucial in life sciences, pharmaceuticals, disease diagnostics, and environmental protection. Infrared spectrum is a powerful tool for identifying different molecules, which can be used to detect the characteristic transmission peaks associated with molecule’s functional groups. However, the complexity of infrared spectrum limits the chirality recognition of amino acid through the traditional analytical techniques. In this work, we propose a method to distinguish three pairs of L– and D–amino acids based on their infrared spectrum by combining machine learning method and sliding window. Three machine learning (ML) algorithms: error-correcting output codes-support vector machine (ECOC-SVM), principal component analysis-random forest (PCA-RF), and partial least squares-discrimination analysis (PLS-DA) are used to recognize three pairs of chiral amino acids including alanine (Ala), cysteine (Cys), and glutamine (Gln), respectively. Results revealed that the sensitivity of chiral recognition varies significantly with spectral region, window size, and step length. Experimental findings highlight the precise chiral recognition of amino acids by ML-assisted infrared spectrum, advancing its applications in analytical chemistry, biomedicine, and pharmaceutical sciences.